Predviđanje cijena dionica zasnovano na modelima dubokog učenja

undergraduate thesis

• Title (english): Prediction of Stock Market Prices Based on Deep Learning Models
• Author: Matej Šikić Vagić
• Mentor: Goran Delač (mentor)
• Committee member: Goran Delač (predsjednik povjerenstva)
• Committee member: Adrian Satja Kurdija (član povjerenstva)
• Committee member: Lucija Šikić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; Zagreb
• Defense date and country: 2020-09-07, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Rad obrađuje temu predviđanje cijena dionica zasnovano na modelima dubokog učenja, točnije koristi se mreža s dugom kratkotrajnom memorijom. Rad će se na početku osvrnuti i na povratne neuronske mreže kao roditelja mreže s dugom kratkotrajnom memorijom. Pomoću mreže s dugom kratkotrajnom memorijom može se predvidjeti rast ili pad neke dionice isključivo s prijašnjim cijenama spomenute dionice. Za ulazni skup informacija uzima se samo datum, cijena dionice na otvaranju i zatvaranju burze. Kasnije, nakon istraživanja trendova, izrađuje se model na setu za treniranje. Nakon raznih preinaka modela uz promjenu raznih parametara poput slučajnog izostavljanja nekih podataka te promjene broja mreža, provjerava se točnost modela na testnom setu podataka koji je odvojen od seta za treniranje.

Abstract (english)

The paper deals with the topic of stock price forecasting based on deep learning models, more precisely using a network with long short-term memory. The paper will initially look at reverse neural networks as the parent of a network with long short-term memory. With the help of a network with long short-term memory, it is possible to predict the growth or decline of a stock only with the previous prices of the mentioned stock. Only the date, share price at the opening and closing of the stock exchange is taken as the input set of information. Later, after researching the trends, a model is made on the training set. After various modifications of the model with a change of various parameters such as accidental omission of some data and a change in the number of networks, the accuracy of the model is checked on a test data set that is separate from the training set.

• Keywords: neuronske mreže; mreže s dugom kratkotrajnom memorijom; Tensorflow; predviđanje cijene dionica; Keras
• Keywords (english): neural networks; long-term memory networks; Tensorflow; stock price prediction; Keras
• Language: croatian
• URN:NBN: urn:nbn:hr:168:391609
• Study programme: Title: Computing; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva (sveučilišni/a prvostupnik/ prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka računarstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2023-01-14 21:45:23